y’all, an LLM aimed hardware Trojan is out. It’s a bit flipper, based on RowHammer, and it can reduce an LLM model from 80% accuracy to .01%. Oh #hackers, you mischievous scamps!
#infosec #llm #ai #hackTheGibson
https://thehackernews.com/2025/07/gpuhammer-new-rowhammer-attack-variant.html?m=1